# GA4, Google Ads and Meta don't match? What's normal

Why the three never agree on sales, what each got when we tested one purchase under four consent choices, and how to tell a normal gap from a broken one.

_Shaun Brown · November 6, 2026 · https://tagsentry.ai/blog/ga4-google-ads-meta-dont-match_

> **The short answer:** They never match, because each one counts different people, puts a sale on a different date and fills gaps with its own modelling. Consent widens the gap: in our test, after a Reject, GA4 and Google Ads still got a cookieless signal for the purchase and Meta's browser pixel got nothing. A gap is normal when it holds steady against your real orders. It means something broke when one platform drops and your orders don't.

## Why the three numbers differ

Each platform answers a different question. GA4 asks "how many purchases happened on the site, and which channel gets the credit?" Google Ads and Meta each ask "how many purchases can we tie to our own ads?"

| | GA4 | Google Ads | Meta Ads Manager |
|---|---|---|---|
| Who it counts | Every visitor it's allowed to measure | People it can tie to a Google ad | People it can tie to a Meta ad |
| Date a sale lands on | The day of the purchase | The day of the ad click, unless you use the "by conv. time" columns | Your choice: the day of the ad or the day of the purchase |
| Window | Looks back 90 days for a purchase by default | A conversion window you set, from 1 to 90 days | 1 or 7 days after a click, 1 day after a view |
| Credit split with other channels | Yes, paid and organic, by default | Google's own ads only, by default | Meta's own ads only |
| Fills gaps | With modelling, once a property is big enough | With modelling, inside the Conversions column | With modelling, and from 12 November by matching sales on the shopper's device |

Two rows explain most month-end arguments. Google Ads files a sale under the day of the click, so a click on the 30th that buys on the 2nd lands in a different month from GA4. And Meta can credit a sale to someone who only saw an ad, where GA4 sees a direct or organic visit.

Meta's help pages give its own reasons for gaps with other tools. It follows people across devices and credits impressions, which pushes its count up. An ad blocker can stop the pixel firing, which pushes it down.

## One purchase, four consent choices

Then there's consent. On 6 November 2026 we built a thank-you page carrying Google's real tag (gtag.js) for a made-up GA4 property and Google Ads account, and Meta's real pixel code (fbevents.js) for a made-up pixel. The page fired one $59 purchase. Every request was recorded and stopped before it left the browser.

We set consent the way a banner would, under four choices. Meta's pixel was set to wait until the visitor agreed, as a tag held by Tag Manager or a banner would.

| Choice | GA4 got | Google Ads got | Meta's pixel got |
|---|---|---|---|
| Accept all | The purchase, with its cookie ID | The conversion, with its cookie ID | The purchase |
| Reject all (a place where the law asks first) | The purchase, cookieless | A cookieless conversion ping | Nothing |
| Banner ignored | The purchase, cookieless | A cookieless conversion ping | Nothing |
| US visitor with GPC (ads opted out, analytics on) | The purchase, with its cookie ID | A cookieless conversion ping | Nothing |

Global Privacy Control (GPC) is a browser setting that tells sites "don't sell or share my data". Our [GPC guide](/blog/global-privacy-control-explained) covers where it counts.

Meta's browser pixel saw one purchase in four. Anyone who said no, or said nothing, is invisible to it unless you also send sales from your server.

GA4 got all four, but a cookieless hit doesn't always become a reported sale. Google models those visitors only once a property sees at least 1,000 events a day from people who declined, for at least 7 days, plus 1,000 daily users who agreed. Below that, Google says purchases from people who declined "aren't reported".

Google Ads got a ping every time. Only the Accept carried Google's ad cookie (`_gcl_au`); the other three were cookieless. Google fills those gaps with modelling once an account has enough ad clicks in a country.

The US visitor with GPC split the platforms. GA4 counted the sale as usual, Meta lost it, and Google Ads got a cookieless ping. If your banner honours GPC, a US store with many GPC visitors will see GA4 run ahead of Meta, and attribution has nothing to do with it.

If your gap appeared the week a banner went live, start with [why analytics dropped after a cookie banner](/blog/why-analytics-dropped-after-cookie-banner), and for Google Ads, [conversions that dropped with Consent Mode](/blog/google-ads-conversions-dropped-consent-mode).

## What a normal gap looks like

A normal gap is steady. Pick a source you trust, usually your store's order count, and track each platform as a share of it, week by week.

GA4 should sit below your orders, by roughly the share of people who decline in ask-first places plus those with ad blockers.

Meta and Google Ads report only the sales they tie to their own ads, so they describe their own reach and were never meant to add up to your sales. Meta's figure can pass what GA4 credits to Meta, thanks to view credit and devices.

None of the three publishes a normal percentage. Your own ratio, and how it moves, is the number to watch.

## Signs it's broken, not normal

- One platform drops and your orders don't. If Meta's purchases halve overnight while orders hold, check the pixel before you blame attribution.
- The ratio jumps on a day you shipped something: a theme update, a checkout app, a new banner or a Tag Manager publish. Our guide to [tags that stop firing](/blog/tags-stop-firing-silently) covers the hunt.
- Purchases arrive but revenue doesn't. A value sent as text, or with no currency, counts as a sale with no money attached. See [GA4 purchase event missing value](/blog/ga4-purchase-event-missing-value).
- A platform reports more purchases than you had orders. In Google Ads that's usually [the same conversion counted twice](/blog/google-ads-conversions-higher-than-actual). Meta suggests its Meta Ads Data Advisor extension to look for duplicate pixel fires, which redirects can cause.

One gap that looks broken and isn't: Meta's Events Manager shows every event it received, while Ads Manager shows only those it ties to an ad. To check the pixel, compare Events Manager's `Purchase` count with your orders.

## Meta's 12 November change

Meta's help pages say "Measure conversions on customer devices" is on by default. It matches an ad view to a purchase on the shopper's own device, using details they already gave you, such as an email address. It isn't active for people in the EU or UK. Trade press reports that eligible pixels start on 12 November.

Meta lists two requirements: the pixel sits directly on the page, and the site requires people to log in with an email or phone number, so a shop with guest checkout may not qualify. Attribution windows stay the same, Meta says.

Where it applies, expect Meta's numbers to rise from 12 November with no change in real sales, about two weeks before Black Friday. Write the date down so nobody reads the jump as a better campaign. The switch is in Events Manager › Data sources › your pixel › Settings.

## Compare like with like

Before you explain any gap to a client, line the reports up:

1. Use the same time zone in all three.
2. In Google Ads, use the "Conversions (by conv. time)" column, which files a sale on the day it happened.
3. In Meta, report by time of conversion, not time of impression.
4. Compare each platform with your orders, not with each other.
5. Check by transaction ID where you can: a sale missing from one platform tells you more than a total.

## Where TagSentry fits

[Event monitoring](/monitoring) watches every destination your tags send to, each on its own. When an event that matters stops arriving at one of them, it emails you once, and once more when it recovers. "Stopped" is judged against your site's own traffic, so a quiet night on a small shop doesn't set it off.

Purchases you've saved as key events are checked for value and currency in what GA4 and Meta each received, once 20 purchases have come in. A sale can leave your site complete and still reach one platform without its value, and that shows up too.

It's one line of code above Google Tag Manager. Purchase checks work on WooCommerce and other checkouts that run your site's own code, once the line is in your site-wide header. In the EU and UK, monitoring counts visitors who allowed analytics.

## Sources

- [Google Ads Help: Create conversions from Google Analytics events (why numbers differ)](https://support.google.com/google-ads/answer/2375435?hl=en)
- [Google Ads Help: Understand conversions by time with new columns](https://support.google.com/google-ads/answer/9549009?hl=en)
- [Analytics Help: Conversions vs. key events in Google Analytics](https://support.google.com/analytics/answer/13965727?hl=en)
- [Analytics Help: Select attribution settings](https://support.google.com/analytics/answer/10597962?hl=en)
- [Analytics Help: Behavioral modeling for consent mode](https://support.google.com/analytics/answer/11161109?hl=en)
- [Google Ads Help: About consent mode modeling](https://support.google.com/google-ads/answer/10548233?hl=en)
- [Meta Business Help: About attribution models and attribution settings](https://www.facebook.com/business/help/2198119873776795)
- [Meta Business Help: Conversion count differences with third-party reporting tools](https://www.facebook.com/business/help/147965221941551)
- [Meta Business Help: Troubleshoot common reporting differences with third-party tools](https://www.facebook.com/business/help/511923449323749)
- [Meta Business Help: About measuring conversions on customer devices](https://www.facebook.com/business/help/4237575203200036)
- [Meta Business Help: How to turn on measuring conversions on customer devices](https://www.facebook.com/business/help/2192875204996675)
- [PPC Land: Meta Pixel will count more conversions from Nov 12 unless advertisers opt out (27 October 2026)](https://ppc.land/meta-pixel-will-count-more-conversions-from-nov-12-unless-advertisers-opt-out/)
